Class

Teaching Computing Methods to First Year Anthropology Students
Michael D. Fischer

University of Kent

BICA Issue No. 6: September 1987

This is an account of the first year of a course I put on at the University of Kent for first-year students in anthropology. The objectives of the course were to provide students background knowledge in applications of computing to social anthropology, give them experience with using some of these methods, and to acquaint them with areas of computer technology where anthropological methods might apply.

Thanks to a Computer Board Teaching Initiative, anthropology at Kent is ideally suited for this course, possessing its own mini-computer and a terminal room exclusively for the use of anthropology students. These students have the best access to computer resources of any first year students on campus.

There was an initial intake of ten, although one withdrew from the university, and one withdrew for medical reasons. There was some initial problem with recruitment for the course, as it was not listed in the course schedule, but this was overcome by other publicity prior to registration.

With one exception, none of the students had any prior knowledge, interest, or experience with computers. The course was broadly divided into three sections. The first section was occupied with basic skills, such as typing, data file manipulations, and operating programs, including text editors, text matchers, and prepared simulations.

The second section consisted of even measures of anthropological theory and methodology, including field methodology, and the principles of translating these into computing terms. Work in this section included simulations, statistics, qualitative analysis, symbolic computing, and applications of expert systems to anthropological analysis. The third section was only six weeks long, and focussed on examining the relationship between society and computing, artificial intelligence, and the use of computers in anthropological fieldwork.

I decided early on not to have a substantial programming component in the course. In retrospect this was an error. There were two sources of problems. First, it was very difficult to explain what the computer was doing in each of the tasks encountered. Although intellectually the students could grasp an explanation, they never got over a sense of magic in practical terms. They could understand the principle, but not the mechanism. Second, the students found it more difficult to become comfortable in the computing environment. Although most were able to use the methods productively for assessments, they had some difficulty with extending the methods to other problem domains. Today's computer is an imperative machine, and the student should have some experience with the mechanics as well as the function.

The First Segment: Skill-Building and Functional Introduction

At the beginning the course focussed on presenting the computer as an artefact of human culture. The idea was to distance the students initially from the machineness of a computer. At the same time students lacked basic skills, principally typing. So I divorced the lectures from the seminars. In the lectures, covered issues included representation and modelling, and a brief account of the intellectual history of computers in their context of use. For the seminars I wrote programs that engaged the students in a dialogue with the computer: they gained familiarity with the machine by cracking simple cryptograms, and by running a model of a society. This stage continued for four weeks, at which time the students could create documents with a text editor, do basic required maintenance functions, and type at a reasonable speed. This time was an orientation, and intended to make the student comfortable, hence proceeding at a leisurely pace. From these relatively clerical beginnings the course moved on to a simple simulation. The students were able to change various ecological and social parameters in the simulation, read the article the simulation was based on, and were asked to write a simple analysis of their findings. The remainder of the term focussed on bibliographic databases, extracting information from data files for reports, and using a database based on Murdock's Atlas of World Cultures . In general, at the end of the term the students felt satisfied with themselves.

The Second Segment: Theory and Method

The second part of the course began with a general history of anthropology, focusing especially on methodology, and its relation to theory. The seminar was spent discussing the possible relevance of computing to different approaches. Then began a two week section on statistics, which grew into a three week section. The emphasis was on appropriate application of programmed procedures, and the interpretation of results. I encountered much more difficulty and resistance than that expected; much of it seemed to stem from a general inability to relate the symbolic aspects of the model underlying univariate statistics to that of the examples. This could be related to math education in the UK. The project for this unit was a series of questions ranging from concrete to abstract drawn from published anthropological work. I was more or less pleased by the answers, especially those on comparing anthropological norms to statistical norms. The course continued with a gentle introduction to graph theory in the form of taxonomic analysis. Realising my mistake with statistics, I developed examples without using the term graph theory until the section was complete. The principal exercises for the students were to classify a domain that was familiar to them, such as flowers, automobiles, or books, into a feature taxonomy, and to classify the links in a social network. The former revealed to students the difficulty in finding a compact feature analysis, and the latter illustrated the problem of representing relative (local) structure and global structure in the same framework. The remainder of the term concentrated on symbolic modelling, with examples based on logical rules, production rules, and phrase-structure rules. The logic rule system is a simple inheritance structure, where categories are created, and objects assigned to a category or categories, inheriting any properties of the categories. This approach in essence maintains multiple taxonomies over a set of objects. Production rules are a common approach to building expert systems. The production rule system was a very simple prototype that only allowed conditions to be true or false. Unlike the logic system, there is no inherent hierarchy built into the production system, that is, the hierarchy changes depending on the available information. The third system is similar to J. Kippen's Bol Machine [(.Kippen Bica#4.)], where a set of generative rules (or a grammar) is created to generate surface structures, and the system can either generate surface structures, or work backwards from surface structures to the rules. Each student was required to model a domain as a class project, using any of the methods. Most of the results were satisfactory.

The Third Segment: Anthropological Components of Computing

The third segment of the course was aimed at applications of anthropology to computing. This was done in three ways. First, examination of the role of computers in contemporary British society, reviewing what computers were used for, and what social changes might be attributed to this use. Second, a discussion of future possibilities of computer use, such as telecommuting, or automated bureaucracy, and what social changes might occur. Third, a case study of artificial intelligence, with emphasis on the ethnocentricity of the goals of AI. In a sense this was closing the circle begun in the first segment; placing computers as artefacts of human culture.

Planned Revisions

The most important change planned is the introduction of symbolic programming early in the first segment, initially using a declarative language I created especially for beginners, and then using a language called Prolog and a production rule language. This is crucial, for it is essential that students gain an early perception of the computer as something that they control, not merely as a machine that `does things'. These languages have the advantage that it is relatively easy to assign interesting problems for the computer with very little initiation. It is hoped that with this experience behind them, they will be able to relate to the ideas underlying more sophisticated applications more easily. The second change will be the movement of the statistics component to the end of the second segment. By beginning with it, the stereotype of social science computing as statistics is reinforced, and even the simple mathematical ideas involved seem to react with the students' mathematical socialisation.

Conclusions

Most of the goals set out for the course were fulfilled. It emerged that perhaps the goals were not sufficient, particularly in the area of giving the student a sense of control over the computer environment, but that will hopefully be improved in the coming year. There was considerable enthusiasm among the students who took the course, although there were some rough spots at times. Out of a class of eight students, two appear quite interested in actively engaging in computer methodology, three were quite pleased with themselves, and three were pleased to finish the course. The situation was essentially ideal, with plenty of access to terminals and a small class, with the exception of a lack of printed reference material. Although there are a number of articles that present research that was accomplished using computer-based methodology, most of these were too anthropologically technical for first-year students. This situation will perhaps improve in coming years.

Return to Contents page